Jun Zhang

Yuting Gao, Jinfeng Liu, Zihan Xu et al. · tencent-ai

Large-scale vision-language pre-training has achieved promising results on downstream tasks. Existing methods highly rely on the assumption that the image-text pairs crawled from the Internet are in perfect one-to-one correspondence. However, in real scenarios, this assumption can be difficult to hold: the text description, obtained by crawling the affiliated metadata of the image, often suffers from the semantic mismatch and the mutual compatibility. To address these issues, we introduce PyramidCLIP, which constructs an input pyramid with different semantic levels for each modality, and aligns visual elements and linguistic elements in the form of hierarchy via peer-level semantics alignment and cross-level relation alignment. Furthermore, we soften the loss of negative samples (unpaired samples) so as to weaken the strict constraint during the pre-training stage, thus mitigating the risk of forcing the model to distinguish compatible negative pairs. Experiments on five downstream tasks demonstrate the effectiveness of the proposed PyramidCLIP. In particular, with the same amount of 15 million pre-training image-text pairs, PyramidCLIP exceeds CLIP on ImageNet zero-shot classification top-1 accuracy by 10.6%/13.2%/10.0% with ResNet50/ViT-B32/ViT-B16 based image encoder respectively. When scaling to larger datasets, PyramidCLIP achieves the state-of-the-art results on several downstream tasks. In particular, the results of PyramidCLIP-ResNet50 trained on 143M image-text pairs surpass that of CLIP using 400M data on ImageNet zero-shot classification task, significantly improving the data efficiency of CLIP.

Feng Luo, Jinxi Xiang, Jun Zhang et al. · tencent-ai, tsinghua

The recent use of diffusion prior, enhanced by pre-trained text-image models, has markedly elevated the performance of image super-resolution (SR). To alleviate the huge computational cost required by pixel-based diffusion SR, latent-based methods utilize a feature encoder to transform the image and then implement the SR image generation in a compact latent space. Nevertheless, there are two major issues that limit the performance of latent-based diffusion. First, the compression of latent space usually causes reconstruction distortion. Second, huge computational cost constrains the parameter scale of the diffusion model. To counteract these issues, we first propose a frequency compensation module that enhances the frequency components from latent space to pixel space. The reconstruction distortion (especially for high-frequency information) can be significantly decreased. Then, we propose to use Sample-Space Mixture of Experts (SS-MoE) to achieve more powerful latent-based SR, which steadily improves the capacity of the model without a significant increase in inference costs. These carefully crafted designs contribute to performance improvements in largely explored 4x blind super-resolution benchmarks and extend to large magnification factors, i.e., 8x image SR benchmarks. The code is available at https://github.com/amandaluof/moe_sr.

Fei Shen, Hu Ye, Jun Zhang et al.

Recent work has showcased the significant potential of diffusion models in pose-guided person image synthesis. However, owing to the inconsistency in pose between the source and target images, synthesizing an image with a distinct pose, relying exclusively on the source image and target pose information, remains a formidable challenge. This paper presents Progressive Conditional Diffusion Models (PCDMs) that incrementally bridge the gap between person images under the target and source poses through three stages. Specifically, in the first stage, we design a simple prior conditional diffusion model that predicts the global features of the target image by mining the global alignment relationship between pose coordinates and image appearance. Then, the second stage establishes a dense correspondence between the source and target images using the global features from the previous stage, and an inpainting conditional diffusion model is proposed to further align and enhance the contextual features, generating a coarse-grained person image. In the third stage, we propose a refining conditional diffusion model to utilize the coarsely generated image from the previous stage as a condition, achieving texture restoration and enhancing fine-detail consistency. The three-stage PCDMs work progressively to generate the final high-quality and high-fidelity synthesized image. Both qualitative and quantitative results demonstrate the consistency and photorealism of our proposed PCDMs under challenging scenarios.The code and model will be available at https://github.com/tencent-ailab/PCDMs.

Chen Shen, Jun Zhang, Xinggong Liang et al.

Forensic pathology is critical in analyzing death manner and time from the microscopic aspect to assist in the establishment of reliable factual bases for criminal investigation. In practice, even the manual differentiation between different postmortem organ tissues is challenging and relies on expertise, considering that changes like putrefaction and autolysis could significantly change typical histopathological appearance. Developing AI-based computational pathology techniques to assist forensic pathologists is practically meaningful, which requires reliable discriminative representation learning to capture tissues' fine-grained postmortem patterns. To this end, we propose a framework called FPath, in which a dedicated self-supervised contrastive learning strategy and a context-aware multiple-instance learning (MIL) block are designed to learn discriminative representations from postmortem histopathological images acquired at varying magnification scales. Our self-supervised learning step leverages multiple complementary contrastive losses and regularization terms to train a double-tier backbone for fine-grained and informative patch/instance embedding. Thereafter, the context-aware MIL adaptively distills from the local instances a holistic bag/image-level representation for the recognition task. On a large-scale database of $19,607$ experimental rat postmortem images and $3,378$ real-world human decedent images, our FPath led to state-of-the-art accuracy and promising cross-domain generalization in recognizing seven different postmortem tissues. The source code will be released on \href{https://github.com/ladderlab-xjtu/forensic_pathology}{https://github.com/ladderlab-xjtu/forensic\_pathology}.

Boxuan Zhao, Jun Zhang, Deheng Ye et al.

Whole-slide images (WSI) in computational pathology have high resolution with gigapixel size, but are generally with sparse regions of interest, which leads to weak diagnostic relevance and data inefficiency for each area in the slide. Most of the existing methods rely on a multiple instance learning framework that requires densely sampling local patches at high magnification. The limitation is evident in the application stage as the heavy computation for extracting patch-level features is inevitable. In this paper, we develop RLogist, a benchmarking deep reinforcement learning (DRL) method for fast observation strategy on WSIs. Imitating the diagnostic logic of human pathologists, our RL agent learns how to find regions of observation value and obtain representative features across multiple resolution levels, without having to analyze each part of the WSI at the high magnification. We benchmark our method on two whole-slide level classification tasks, including detection of metastases in WSIs of lymph node sections, and subtyping of lung cancer. Experimental results demonstrate that RLogist achieves competitive classification performance compared to typical multiple instance learning algorithms, while having a significantly short observation path. In addition, the observation path given by RLogist provides good decision-making interpretability, and its ability of reading path navigation can potentially be used by pathologists for educational/assistive purposes. Our code is available at: \url{https://github.com/tencent-ailab/RLogist}.

Silvio Giancola, Anthony Cioppa, Adrien Deliège et al.

The SoccerNet 2022 challenges were the second annual video understanding challenges organized by the SoccerNet team. In 2022, the challenges were composed of 6 vision-based tasks: (1) action spotting, focusing on retrieving action timestamps in long untrimmed videos, (2) replay grounding, focusing on retrieving the live moment of an action shown in a replay, (3) pitch localization, focusing on detecting line and goal part elements, (4) camera calibration, dedicated to retrieving the intrinsic and extrinsic camera parameters, (5) player re-identification, focusing on retrieving the same players across multiple views, and (6) multiple object tracking, focusing on tracking players and the ball through unedited video streams. Compared to last year's challenges, tasks (1-2) had their evaluation metrics redefined to consider tighter temporal accuracies, and tasks (3-6) were novel, including their underlying data and annotations. More information on the tasks, challenges and leaderboards are available on https://www.soccer-net.org. Baselines and development kits are available on https://github.com/SoccerNet.

Siteng Chen, Jinxi Xiang, Xiyue Wang et al.

Tumor mutational burden (TMB) is a potential genomic biomarker of immunotherapy. However, TMB detected through whole exome sequencing lacks clinical penetration in low-resource settings. In this study, we proposed a multi-scale deep learning framework to address the detection of TMB status from routinely used whole slide images for a multiple cancer TMB prediction model (MC- TMB). The MC-TMB achieved a mean area under the curve (AUC) of 0.818 (0.804-0.831) in the cross-validation cohort, which showed superior performance to each single-scale model. The improvements of MC-TMB over the single-tumor models were also confirmed by the ablation tests on x10 magnification, and the highly concerned regions typically correspond to dense lymphocytic infiltration and heteromorphic tumor cells. MC-TMB algorithm also exhibited good generalization on the external validation cohort with an AUC of 0.732 (0.683-0.761), and better performance when compared to other methods. In conclusion, we proposed a deep learning-based approach to reveal tumor mutational burden status from routinely used pathological slides across multiple cancer types.

Fei Kong, Xiyue Wang, Jinxi Xiang et al.

Artificial intelligence (AI) holds significant promise in transforming medical imaging, enhancing diagnostics, and refining treatment strategies. However, the reliance on extensive multicenter datasets for training AI models poses challenges due to privacy concerns. Federated learning provides a solution by facilitating collaborative model training across multiple centers without sharing raw data. This study introduces a federated attention-consistent learning (FACL) framework to address challenges associated with large-scale pathological images and data heterogeneity. FACL enhances model generalization by maximizing attention consistency between local clients and the server model. To ensure privacy and validate robustness, we incorporated differential privacy by introducing noise during parameter transfer. We assessed the effectiveness of FACL in cancer diagnosis and Gleason grading tasks using 19,461 whole-slide images of prostate cancer from multiple centers. In the diagnosis task, FACL achieved an area under the curve (AUC) of 0.9718, outperforming seven centers with an average AUC of 0.9499 when categories are relatively balanced. For the Gleason grading task, FACL attained a Kappa score of 0.8463, surpassing the average Kappa score of 0.7379 from six centers. In conclusion, FACL offers a robust, accurate, and cost-effective AI training model for prostate cancer pathology while maintaining effective data safeguards.

Zeqiang Lai, Ying Fu, Jun Zhang

Fusion-based hyperspectral image (HSI) super-resolution has become increasingly prevalent for its capability to integrate high-frequency spatial information from the paired high-resolution (HR) RGB reference image. However, most of the existing methods either heavily rely on the accurate alignment between low-resolution (LR) HSIs and RGB images, or can only deal with simulated unaligned RGB images generated by rigid geometric transformations, which weakens their effectiveness for real scenes. In this paper, we explore the fusion-based HSI super-resolution with real RGB reference images that have both rigid and non-rigid misalignments. To properly address the limitations of existing methods for unaligned reference images, we propose an HSI fusion network with heterogenous feature extractions, multi-stage feature alignments, and attentive feature fusion. Specifically, our network first transforms the input HSI and RGB images into two sets of multi-scale features with an HSI encoder and an RGB encoder, respectively. The features of RGB reference images are then processed by a multi-stage alignment module to explicitly align the features of RGB reference with the LR HSI. Finally, the aligned features of RGB reference are further adjusted by an adaptive attention module to focus more on discriminative regions before sending them to the fusion decoder to generate the reconstructed HR HSI. Additionally, we collect a real-world HSI fusion dataset, consisting of paired HSI and unaligned RGB reference, to support the evaluation of the proposed model for real scenes. Extensive experiments are conducted on both simulated and our real-world datasets, and it shows that our method obtains a clear improvement over existing single-image and fusion-based super-resolution methods on quantitative assessment as well as visual comparison.

Kunyuan Li, Jun Zhang, Jun Gao et al.

Exploiting light field data makes it possible to obtain dense and accurate depth map. However, synthetic scenes with limited disparity range cannot contain the diversity of real scenes. By training in synthetic data, current learning-based methods do not perform well in real scenes. In this paper, we propose a self-supervised learning framework for light field depth estimation. Different from the existing end-to-end training methods using disparity label per pixel, our approach implements network training by estimating EPI disparity shift after refocusing, which extends the disparity range of epipolar lines. To reduce the sensitivity of EPI to noise, we propose a new input mode called EPI-Stack, which stacks EPIs in the view dimension. This method is less sensitive to noise scenes than traditional input mode and improves the efficiency of estimation. Compared with other state-of-the-art methods, the proposed method can also obtain higher quality results in real-world scenarios, especially in the complex occlusion and depth discontinuity.

Kuan Tian, Yonghang Guan, Jinxi Xiang et al.

The state-of-the-art neural video codecs have outperformed the most sophisticated traditional codecs in terms of RD performance in certain cases. However, utilizing them for practical applications is still challenging for two major reasons. 1) Cross-platform computational errors resulting from floating point operations can lead to inaccurate decoding of the bitstream. 2) The high computational complexity of the encoding and decoding process poses a challenge in achieving real-time performance. In this paper, we propose a real-time cross-platform neural video codec, which is capable of efficiently decoding of 720P video bitstream from other encoding platforms on a consumer-grade GPU. First, to solve the problem of inconsistency of codec caused by the uncertainty of floating point calculations across platforms, we design a calibration transmitting system to guarantee the consistent quantization of entropy parameters between the encoding and decoding stages. The parameters that may have transboundary quantization between encoding and decoding are identified in the encoding stage, and their coordinates will be delivered by auxiliary transmitted bitstream. By doing so, these inconsistent parameters can be processed properly in the decoding stage. Furthermore, to reduce the bitrate of the auxiliary bitstream, we rectify the distribution of entropy parameters using a piecewise Gaussian constraint. Second, to match the computational limitations on the decoding side for real-time video codec, we design a lightweight model. A series of efficiency techniques enable our model to achieve 25 FPS decoding speed on NVIDIA RTX 2080 GPU. Experimental results demonstrate that our model can achieve real-time decoding of 720P videos while encoding on another platform. Furthermore, the real-time model brings up to a maximum of 24.2\% BD-rate improvement from the perspective of PSNR with the anchor H.265.

Kuan Tian, Yonghang Guan, Jinxi Xiang et al.

Under certain circumstances, advanced neural video codecs can surpass the most complex traditional codecs in their rate-distortion (RD) performance. One of the main reasons for the high performance of existing neural video codecs is the use of the entropy model, which can provide more accurate probability distribution estimations for compressing the latents. This also implies the rigorous requirement that entropy models running on different platforms should use consistent distribution estimations. However, in cross-platform scenarios, entropy models running on different platforms usually yield inconsistent probability distribution estimations due to floating point computation errors that are platform-dependent, which can cause the decoding side to fail in correctly decoding the compressed bitstream sent by the encoding side. In this paper, we propose a cross-platform video compression framework based on codebooks, which avoids autoregressive entropy modeling and achieves video compression by transmitting the index sequence of the codebooks. Moreover, instead of using optical flow for context alignment, we propose to use the conditional cross-attention module to obtain the context between frames. Due to the absence of autoregressive modeling and optical flow alignment, we can design an extremely minimalist framework that can greatly benefit computational efficiency. Importantly, our framework no longer contains any distribution estimation modules for entropy modeling, and thus computations across platforms are not necessarily consistent. Experimental results show that our method can outperform the traditional H.265 (medium) even without any entropy constraints, while achieving the cross-platform property intrinsically.

Feng Ding, Jun Zhang, Xinan He et al.

The high-quality, realistic images generated by generative models pose significant challenges for exposing them.So far, data-driven deep neural networks have been justified as the most efficient forensics tools for the challenges. However, they may be over-fitted to certain semantics, resulting in considerable inconsistency in detection performance across different contents of generated samples. It could be regarded as an issue of detection fairness. In this paper, we propose a novel framework named Fairadapter to tackle the issue. In comparison with existing state-of-the-art methods, our model achieves improved fairness performance. Our project: https://github.com/AppleDogDog/FairnessDetection

Kunyuan Li, Jun Zhang, Rui Sun et al.

Light field cameras record not only the spatial information of observed scenes but also the directions of all incoming light rays. The spatial and angular information implicitly contain geometrical characteristics such as multi-view or epipolar geometry, which can be exploited to improve the performance of depth estimation. An Epipolar Plane Image (EPI), the unique 2D spatial-angular slice of the light field, contains patterns of oriented lines. The slope of these lines is associated with the disparity. Benefiting from this property of EPIs, some representative methods estimate depth maps by analyzing the disparity of each line in EPIs. However, these methods often extract the optimal slope of the lines from EPIs while ignoring the relationship between neighboring pixels, which leads to inaccurate depth map predictions. Based on the observation that an oriented line and its neighboring pixels in an EPI share a similar linear structure, we propose an end-to-end fully convolutional network (FCN) to estimate the depth value of the intersection point on the horizontal and vertical EPIs. Specifically, we present a new feature-extraction module, called Oriented Relation Module (ORM), that constructs the relationship between the line orientations. To facilitate training, we also propose a refocusing-based data augmentation method to obtain different slopes from EPIs of the same scene point. Extensive experiments verify the efficacy of learning relations and show that our approach is competitive to other state-of-the-art methods. The code and the trained models are available at https://github.com/lkyahpu/EPI_ORM.git.

Chuang Niu, Jun Zhang, Ge Wang et al.

We propose a self-supervised Gaussian ATtention network for image Clustering (GATCluster). Rather than extracting intermediate features first and then performing the traditional clustering algorithm, GATCluster directly outputs semantic cluster labels without further post-processing. Theoretically, we give a Label Feature Theorem to guarantee the learned features are one-hot encoded vectors, and the trivial solutions are avoided. To train the GATCluster in a completely unsupervised manner, we design four self-learning tasks with the constraints of transformation invariance, separability maximization, entropy analysis, and attention mapping. Specifically, the transformation invariance and separability maximization tasks learn the relationships between sample pairs. The entropy analysis task aims to avoid trivial solutions. To capture the object-oriented semantics, we design a self-supervised attention mechanism that includes a parameterized attention module and a soft-attention loss. All the guiding signals for clustering are self-generated during the training process. Moreover, we develop a two-step learning algorithm that is memory-efficient for clustering large-size images. Extensive experiments demonstrate the superiority of our proposed method in comparison with the state-of-the-art image clustering benchmarks. Our code has been made publicly available at https://github.com/niuchuangnn/GATCluster.

Sen Yang, Feng Luo, Jun Zhang et al.

Mitotic count is the most important morphological feature of breast cancer grading. Many deep learning-based methods have been proposed but suffer from domain shift. In this work, we construct a Fourier-based segmentation model for mitosis detection to address the problem. Swapping the low-frequency spectrum of source and target images is shown effective to alleviate the discrepancy between different scanners. Our Fourier-based segmentation method can achieve F1 with 0.7456 on the preliminary test set.

Jun Zhang, Kuan Tian, Pei Dong et al.

The overexpression of human epidermal growth factor receptor 2 (HER2) has been established as a therapeutic target in multiple types of cancers, such as breast and gastric cancers. Immunohistochemistry (IHC) is employed as a basic HER2 test to identify the HER2-positive, borderline, and HER2-negative patients. However, the reliability and accuracy of HER2 scoring are affected by many factors, such as pathologists' experience. Recently, artificial intelligence (AI) has been used in various disease diagnosis to improve diagnostic accuracy and reliability, but the interpretation of diagnosis results is still an open problem. In this paper, we propose a real-time HER2 scoring system, which follows the HER2 scoring guidelines to complete the diagnosis, and thus each step is explainable. Unlike the previous scoring systems based on whole-slide imaging, our HER2 scoring system is integrated into an augmented reality (AR) microscope that can feedback AI results to the pathologists while reading the slide. The pathologists can help select informative fields of view (FOVs), avoiding the confounding regions, such as DCIS. Importantly, we illustrate the intermediate results with membrane staining condition and cell classification results, making it possible to evaluate the reliability of the diagnostic results. Also, we support the interactive modification of selecting regions-of-interest, making our system more flexible in clinical practice. The collaboration of AI and pathologists can significantly improve the robustness of our system. We evaluate our system with 285 breast IHC HER2 slides, and the classification accuracy of 95\% shows the effectiveness of our HER2 scoring system.

Jun Zhang, Mina Henein, Robert Mahony et al.

The problem of tracking self-motion as well as motion of objects in the scene using information from a camera is known as multi-body visual odometry and is a challenging task. This paper proposes a robust solution to achieve accurate estimation and consistent track-ability for dynamic multi-body visual odometry. A compact and effective framework is proposed leveraging recent advances in semantic instance-level segmentation and accurate optical flow estimation. A novel formulation, jointly optimizing SE(3) motion and optical flow is introduced that improves the quality of the tracked points and the motion estimation accuracy. The proposed approach is evaluated on the virtual KITTI Dataset and tested on the real KITTI Dataset, demonstrating its applicability to autonomous driving applications. For the benefit of the community, we make the source code public.

Yinghui Zhang, Xiaojuan Deng, Jun Zhang et al.

Euler's elastica model has been extensively studied and applied to image processing tasks. However, due to the high nonlinearity and nonconvexity of the involved curvature term, conventional algorithms suffer from slow convergence and high computational cost. Various fast algorithms have been proposed, among which, the augmented Lagrangian based ones are very popular in the community. However, parameter tuning might be very challenging for these methods. In this paper, a simple cutting-off strategy is introduced into the augmented Lagrangian based algorithms for minimizing the Euler's elastica energy, which leads to easy parameter tuning and fast convergence. The cutting-off strategy is based on an observation of inconsistency inside the augmented Lagrangian based algorithms. When the weighting parameter of the curvature term goes to zero, the energy functional boils down to the ROF model. So, a natural requirement is that its augmented Lagrangian based algorithms should also approach the augmented Lagrangian based algorithms formulated directly for solving the ROF model from the very beginning. Unfortunately, this is not the case for certain existing augmented Lagrangian based algorithms. The proposed cutting-off strategy helps to decouple the tricky dependence between the auxiliary splitting variables, so as to remove the observed inconsistency. Numerical experiments suggest that the proposed algorithm enjoys easier parameter-tuning, faster convergence and even higher quality of image restorations.

Jun Zhang

Deformable image registration is a fundamental task in medical image analysis, aiming to establish a dense and non-linear correspondence between a pair of images. Previous deep-learning studies usually employ supervised neural networks to directly learn the spatial transformation from one image to another, requiring task-specific ground-truth registration for model training. Due to the difficulty in collecting precise ground-truth registration, implementation of these supervised methods is practically challenging. Although several unsupervised networks have been recently developed, these methods usually ignore the inherent inverse-consistent property (essential for diffeomorphic mapping) of transformations between a pair of images. Also, existing approaches usually encourage the to-be-estimated transformation to be locally smooth via a smoothness constraint only, which could not completely avoid folding in the resulting transformation. To this end, we propose an Inverse-Consistent deep Network (ICNet) for unsupervised deformable image registration. Specifically, we develop an inverse-consistent constraint to encourage that a pair of images are symmetrically deformed toward one another, until both warped images are matched. Besides using the conventional smoothness constraint, we also propose an anti-folding constraint to further avoid folding in the transformation. The proposed method does not require any supervision information, while encouraging the diffeomoprhic property of the transformation via the proposed inverse-consistent and anti-folding constraints. We evaluate our method on T1-weighted brain magnetic resonance imaging (MRI) scans for tissue segmentation and anatomical landmark detection, with results demonstrating the superior performance of our ICNet over several state-of-the-art approaches for deformable image registration. Our code will be made publicly available.

Zhe Zhu, Michael Harowicz, Jun Zhang et al.

Purpose: To determine whether deep learning-based algorithms applied to breast MR images can aid in the prediction of occult invasive disease following the di- agnosis of ductal carcinoma in situ (DCIS) by core needle biopsy. Material and Methods: In this institutional review board-approved study, we analyzed dynamic contrast-enhanced fat-saturated T1-weighted MRI sequences of 131 patients at our institution with a core needle biopsy-confirmed diagnosis of DCIS. The patients had no preoperative therapy before breast MRI and no prior history of breast cancer. We explored two different deep learning approaches to predict whether there was a hidden (occult) invasive component in the analyzed tumors that was ultimately detected at surgical excision. In the first approach, we adopted the transfer learning strategy, in which a network pre-trained on a large dataset of natural images is fine-tuned with our DCIS images. Specifically, we used the GoogleNet model pre-trained on the ImageNet dataset. In the second approach, we used a pre-trained network to extract deep features, and a support vector machine (SVM) that utilizes these features to predict the upstaging of the DCIS. We used 10-fold cross validation and the area under the ROC curve (AUC) to estimate the performance of the predictive models. Results: The best classification performance was obtained using the deep features approach with GoogleNet model pre-trained on ImageNet as the feature extractor and a polynomial kernel SVM used as the classifier (AUC = 0.70, 95% CI: 0.58- 0.79). For the transfer learning based approach, the highest AUC obtained was 0.53 (95% CI: 0.41-0.62). Conclusion: Convolutional neural networks could potentially be used to identify occult invasive disease in patients diagnosed with DCIS at the initial core needle biopsy.